Saturday, August 31, 2013

More interesting USTA League stats - Which court does the winning team win most often?

Continuing on in my effort to share interesting stats from the data I collect for the Estimated Dynamic NTRP Rating reports I generate, I took a look at how often the winning team in a match wins the different courts.

Specifically, I took a look at all team matches that were 2 singles and 3 doubles.  Here is the percent of the time the winning team wins each court.

CourtWinning %
1S72.9
2S73.8
1D74.9
2D76.2
3D75.5

It isn't a huge margin, but pretty clearly the winning team is going to win court 2 doubles more often than any other court.  And court 1 singles is won the least often.  And interestingly, all of the doubles courts are won more often than either of the singles courts.

What can we take away from this?

First, it appears that the key to winning is having good doubles teams and not singles guys.  This goes against intuition a bit (IMHO) that the easiest way to having a strong team is to have 2 lock-down singles guys and 1 great doubles team (this requires the fewest strong players to win a team match, just 4).

Second, you need depth at doubles.  Whether it is because captains flip-flop and play their best doubles team on court 2, or just because the better team is deeper and their second best doubles team is better than the opponent's, the winning team wins court 2 doubles most often and court 3 doubles next most often.

Third, and this one probably isn't a surprise, it looks like the most likely court to be thrown when juggling line-ups is court 1 singles with the winning team losing it more often than court 2 singles.  And court 1 doubles also suffers from this too.

What do you think?  What other interesting stats are you interested in seeing?  Let me know.

Tuesday, August 27, 2013

Early Start Leagues and USTA Nationals, are they fair?

I was contacted by a report customer today asking why her 3.0 team was going to have to play against a 3.5 at Nationals, and not just a 3.5 but one that had won a match playing up at 4.0.  The answer is, Early Start Leagues.

In this case, the player in question played in a district where the 40 & over league was an Early Start League this year.  This means that, typically for court scheduling and the like, the 2013 league started play before year-end ratings came out and so it is designated as an Early Start League (ESL).  Because the year-end ratings haven't come out yet, their year-end rating can't be used to determine the level they are eligible to play at.

What sections do to address this is to have Early Start Ratings (ESR) that come out mid-year, typically in July or August.  These ratings are a "sneak preview" into what a players year-end rating may be and are used to try to get the players in an ESL playing at the right level.  This is a good idea and usually works, but sometimes it doesn't.

In this case, the player was a 3.0 in 2012 and played up at 3.5 losing all her matches prior to the ESR date and her dynamic rating was just below 3.0, so her ESR was still 3.0 and she proceeded to begin play in the ESL on a 3.0 team.  She played one match before year-end ratings came out and it was a thumping of the opponent which caused her rating to go up and she ended up being a 3.5 in the year-end ratings.  Note that my ratings correctly had her a 3.0 ESR and also had her being bumped up at year-end.

So she is now a 3.5, but playing in a 3.0 flight.  What are the USTA's rules for this?  See page 12 at this link, but the summary is that sections have options, and it appears option 2 is employed by the section in question which allows a player to continue play in the ESL using their ESR unless their year-end rating was clearly above level.  In this case, the player had just eked over the threshold to be bumped up so she wasn't clearly above level.

So she gets to play in the 3.0 flight, but must now play as a 3.5 in the 18 & over league (not an ESL), and also decides to play up in a 4.0 18 & over flight.  She does get a win at 4.0 (4 losses too), and also loses 3 3.5 matches, so she isn't dominating, but her 3.0 team advances to Nationals.

To opposing teams then, it looks like she is a 3.5 that has won at 4.0 and gets to play against 3.0s at Nationals.  This is all per the rules so is fair in that regard, and many of the other players that will be at Nationals are also effectively 3.5s so the play may be competitive, but should the USTA allow this to happen?

Comments welcome.

Monday, August 26, 2013

More teams advance to USTA Nationals with the help of Estimated Dynamic NTRP Rating reports

USTA League Sectionals took place this past weekend in many areas and I'm pleased to say that two more teams I helped with individual and/or team Estimated Dynamic NTRP Rating reports have advanced to Nationals.  That brings the tally to four.  Thankfully, all four are at different levels or different divisions so they won't be facing each other :)

Unfortunately, one other team I was helping did not advance, losing on the sets lost tie-breaker to a team they beat.  The team I was helping, team A, went 1-1 in their first two matches, their loss being a heart breaking 3-2 loss where all the losses were in match tie-breaks, and they needed to beat the lone remaining 2-0 team in their sub-flight 4-1, or a 3-2 win with a little luck would do it too.  Unfortunately, they won 3-2 but lost a set in two of the wins so they lost more sets and lost that tie-breaker.

This isn't the first time a team has won the head-to-head but lost this tie-breaker, but perhaps it will be the last?  I'm was told that a USTA official said next year, head-to-head will be the first tie-breaker.  Too bad it wasn't this year!

Good luck teams!

Friday, August 23, 2013

More Sectionals take place this weekend

It is August, so there are Sectionals going on somewhere most every weekend.  This weekend has 18 & Over Sectionals going on in Northern California and 40 & Over Sectionals in the Pacific Northwest.  Alas, I was hoping to be playing this weekend but my team lost in our local playoffs and didn't make it to Yakima.

However, I'm still involved as I have teams that have gotten reports participating in both of these Sectionals.  Good luck to those teams!

Thursday, August 15, 2013

Updated interesting USTA League Stats - Ratings by Section - Florida and Midwest added

I've written a few blog posts with interesting ratings, and below is an updated chart showing the ratings by section with Florida and the Midwest added.

NTRP Ratings by Section

Florida is similar to the Mid-Atlantic with a pretty good distribution towards the higher rated players, but is not to the same levels that Northern Cal, Southern Cal, and Texas are.  The Midwest is also very similar to Florida and Mid-Atlantic.

Enjoy!

Wednesday, August 14, 2013

Can a 3.5 USTA League team make it to 3.5 Nationals?

The subject may seem a bit odd, but there is a common perception that a team of players that are actually 3.5s has no chance of advancing to Nationals.  Instead a team must be filled with ringers and self-rated players that under rate themselves and manage their to avoid strikes.

To test this, I thought I'd take a look at a couple 3.5 teams headed to this year's Nationals.  The teams I'm looking at are from Houston, an area notorious for ratings manipulation shenanigans, and the Northwest Washington district that has had a strong record of sending competitive teams to Nationals.  I'll be using my Estimated Dynamic NTRP Ratings to do the analysis.

First, Houston.  This team has a big roster with 7 benchmark rated players and a whopping 14 self-rated or appeal players.  They won their local league with a 7-1 record, winning 32 of 40 courts during the year and then turned it on in local playoffs going 3-0/14-1.  At Sectionals they went 3-0/12-3 winning their flight and then 2-0/7-3 in the semis and finals.

They did have some attrition along the way with 3 of the self-rates being DQ'd, one at Sectionals.  Note that my ratings agree with 2 of them including the one at Sectionals, and perhaps the other was an administrative DQ.  Including those DQ'd players the average rating for the team is 3.51.  But even without them the average is still 3.46.  With the players still eligible for Nationals, they could run out 5 courts with players all rated 3.48 or higher.

This is certainly a strong team with 3 that have already effectively been bumped up and a number of others (not withstanding results of the Texas Fall league being used to manage ratings back down) that will follow suit at the end of the year.  One can certainly make the case that very few of these players are "real" 3.5s.

Next, Northwest Washington.  This team went 8-1/35-10 in league play, going 2-0/9-1 in local playoffs.  At Sectionals, they went 2-0/8-2 in their flight and then won the final 3-2.  Their roster includes 5 self-rates and 4 benchmark players.

None of their self-rates were DQ'd, although 2 of them I do have rated over 3.50.  Their team average rating is just 3.35, but their top-8 average is 3.49 and all are at 3.35 or above.

This team also appears strong, but perhaps not as strong as Houston, and has a number of players that will likely be bumped up at year-end.

Having said all this, you would expect teams that do well and advance to Nationals to have strong players at the top of their rating range and even above.  Players certainly improve and someone has to be bumped up, and you would expect players on teams that do well to fall into this category.

So technically, both teams are filled with 3.5s that are eligible to play.  On the surface it does appear some players, more so from the Houston team, self-rated artificially low and thus aren't really 3.5s, but from what I've presented, I can't say whether this was deliberate or innocent, but it is a shame if it is the former.

It will be interesting to see how these teams do at Nationals, and even more interesting if they happen to face each other.  Check back in a couple months and I'll take a look at how each does at Nationals!

Monday, August 12, 2013

Comparing an Estimated Dynamic NTRP Rating Report with a real USTA League DQ letter

I created some Estimated Dynamic NTRP Rating Reports for a team last week and one of their players had been three-strike disqualified so I had the opportunity to take a look at it.  This isn't new, I've done this many times and generally find that I can explain the DQ.

What was interesting about this one was that the DQ letter they forwarded to me not only identified which matches were strikes, but listed the dynamic rating for each match.  This was a huge opportunity for me to do a more detailed comparison and here is what I found.

The player in question had self-rated as a 4.0 and played 7 matches, all doubles, going 6-1.  Here are the scores, the dynamic rating in the DQ letter, and my estimated rating after each match.

Match ResultDynamic
NTRP
Estimated
Dynamic
NTRP
Won 7-5,0-6,1-03.974.01
Won 6-4,6-14.364.49
Won 7-5,6-14.234.22
Won 6-2,7-64.164.15
Lost 6-3,6-24.174.17
Won 6-2,4-6,1-04.194.17
Won 6-4,7-54.204.20

While my estimates weren't perfect for all 7 matches, they were close, and from the 3rd match on no more than 0.02 off and exact on 2 of the ratings.  I think that is accuracy I can live with.

If you received a DQ letter that included your ratings like this player did, please contact me and I'll generate a report for you for free comparing my estimates with the actual ratings and helping you understand why the ratings were what they were.  You may also contact me if you are just interested in purchasing an individual or team report as well.

A Northern California team report customer advances to USTA Sectionals

This weekend was a busy one for USTA League playoffs and in addition to the Pacific Northwest team that won their section and is headed to Nationals, I helped a team in Northern California that won Districts and is headed to their Sectionals in a couple weeks.  This team is already busy preparing, and I just generated reports for all the teams they may face there.  The teams look very close, so the match-ups they can get with the help of the reports may make the difference for them.

Sunday, August 11, 2013

Another report customer advances to Nationals

This one was inevitable no matter who won as I'd done reports for both teams, but another team I've done reports for, this one from the Pacific Northwest Section, is headed to Nationals.

As I've done before with some playoff match-ups, I took a look at  how my ratings did predicting games.  Here is how my ratings did predicting the matches in the final:
  • A 3.64 beat a 3.51 7-6,6-0
  • A 3.63 beat a 3.50 6-0,6-4
  • A 3.48/3.47 beat a 3.26/3.28 6-2,4-6,1-0
  • A 3.49/3.63 beat a 3.37/3.83 6-1,6-3
  • A 3.28/3.31 beat a 3.42/3.39 3-6,7-6,1-0
Technically, my predictions went 3-2, but the 2 losses if not expected, are explainable.

Any time doubles parters are nearly a full rating apart, it can be an easy target for the opponents to pick on the weaker player and the stronger player can't get involved.  This would explain the 4th match above.

And while the 5th match above was an incorrect pick, the stronger team did win more games.  Unfortunately for the losing team who lost 3-2, they had the on paper stronger team that lost this 5th match.

Good luck at Nationals guys.

Saturday, August 10, 2013

More teams advance in USTA playoffs with the help of Estimated Dynamic NTRP Rating Reports

The Pacific Northwest Section of the USTA is holding their Sectionals for the 18 & over league, and I just learned that two teams I've done individual and team reports for are going to face off in the final.  I can't take credit for them making it this far, they've had to play well on the court, but they've both said the reports helped them prepare and know what to expect going in to the matches.

The good news is, one of them is going to Nationals, joining another team I helped from Northern California as teams making it to their respective Nationals.  And others I've helped still have their Sectionals to play.

Good luck!

Thursday, August 8, 2013

Golf has handicaps, tennis has NTRP ratings

In any recreational sport, there are players with varying degrees of skill, the full range from beginner to aspiring or former professional.

In golf there is a handicapping system where by a player enters their scores and the course/tees played into the system and a handicap is calculated.  This number, very roughly how many shots over par a player may be expected to shoot on a regular basis, allows players of different ability to compete on an equitable basis.

For example, a player A with a handicap of 10 plays player B with a handicap of 6, player A will "get 4 strokes" in order to be able to equitably compare scores at the end of the round.  In theory, two players of any skill set can play and have the playing field leveled through the use of their handicaps.

In tennis, we don't have a way for a near pro to play a beginner and have a level playing field, but we do have the NTRP rating system used by the USTA to rate players at several different standard levels, so that players of the same level can play and have competitive matches.

Of course, no system is foolproof or safe from being manipulated, and this happens with both golf and tennis.  Some golfers choose not to enter good rounds because it would lower their handicap and they'd get fewer strokes, and some tennis players choose to throw matches to lower their rating so they can play at a lower level against weaker competition just so they can win.  The fact that either of these occurs is unfortunate.

In golf, the handicapping system has some provisions built in to try to combat manipulation.  For one, only your 10 best rounds out of your last 20 count towards your handicap.  So unless you are going to throw a lot of rounds, the bad ones don't count and only your good rounds that reflect your capabilities will go into the calculations.

With the NTRP system, to my knowledge, there is no such omitting of results, but perhaps there should be.  Further, some sections/districts choose to include or not various leagues and tournaments.  This gives some players ample opportunity to use "secondary" leagues to throw matches and lower their ratings.  If I were doing the system myself, I'd probably build in some checks to preclude counting thrown matches or throw out highs and lows in the calculations.

To be fair, the NTRP system does have year-end calculations where different matches are weighted more heavily, specifically those in playoffs and those played against players who played in the post-season.  By doing this, the matches that players are likely trying their best in count more which makes sense.

And this is the time of year that these post-season matches are taking place, so these are the ones that will count the most.  Good luck to all those playing in playoffs.  Go out there and play your best and if you get bumped up, congratulations and embrace the challenge of playing tougher opponents next year.

Tuesday, August 6, 2013

Have you been 3 strike disqualified in USTA League tennis? Get a free report that explains the DQ

Analyzing USTA League disqualifications is one of the interesting things I get to do with my Estimated Dynamic NTRP Ratings.  I'm usually able to predict or explain most DQs which is a confirmation that my algorithm is pretty darn close to estimating ratings correctly.

But when I can't explain one, the details of the DQ allow me to look at my algorithm and figure out how I can fine tune it a bit more.  Because this review is so valuable to me, I'm offering a free Estimated Dynamic NTRP Rating Report to the first 5 USTA League players to contact me that received a DQ during the 2013 season.  In order to qualify for the free report, you'll need to send me the actual DQ letter you received with the details about which matches resulted in strikes.

Thanks!

Update: I've made the offer again for 2014.

Sunday, August 4, 2013

More interesting USTA League statistics - Ratings by Section

I wrote yesterday about some interesting statistics I can look at based on the data I analyze for my Estimated Dynamic NTRP Rating Reports.  I took a deeper look and broke out the percentage at each level by the sections that I have a reasonable amount of data for.

NTRP Ratings by Section

This gives an interesting picture of the distribution of players in each section.

Northern Cal, So. Cal, and Texas clearly have the fewest 2.5s and below and not surprisingly, they have the most 4.5s and above.  The number of 3.5s is interesting similar across all of the sections.

This analysis started in response to a dialog about Atlanta vs NorCal and high rated players, and while this simply includes Atlanta in the Southern section, but this shows a higher percentage of strong players in NorCal than Southern.  Unfortunately for NorCal, it appears SoCal has you beat slightly in that category.

Note that this analysis is based on data that numbers in the thousands of players in each section listed above, but not necessarily every player.  And I left out some sections where I have data but the count is smaller and may not be fully representative.

Saturday, August 3, 2013

Some interesting USTA League stats - Players by NTRP level

As a by product of generating individual and team Estimated Dynamic NTRP Rating reports, I have a fair amount of data from which I can calculate some interesting statistics.

A Talk Tennis thread on sandbagging turned into a discussion about how Atlanta and the San Francisco Bay Area compared and I posted some stats on how many players there were by level from each area.  I thought it would be interested to take a look at the same stats for all the data I have and here it is.

  • 2.5- - 5%
  • 3.0 - 21%
  • 3.5 - 36%
  • 4.0 - 28%
  • 4.5 - 9%
  • 5.0+ - 1%

USTA League Players by Rating (%)

These numbers are probably not a surprise, but it is still interesting to see them for real.  The 3.5 level has the most players followed by 4.0 and then 3.0.

Do you have another stat you'd like me to post?  Leave a comment and let me know.

Note: These stats are for about half the sections in the USTA but are from a large enough data set that they are probably representative of USTA participation as a whole.

August is the month for USTA Sectionals

For teams looking to get to USTA League Nationals, August is an important month.  With the realignment of the leagues to add the 40 & over league in addition the 18 & over league, some sections may have had playoffs/districts/sectionals for one of the leagues already, but for most, August and September is the time that teams advancing to Sectionals and Nationals are determined.

Where I'm at in the Pacific Northwest section, August is when our sectionals take place.  The 18 & over league will be played August 9-11 in Portland, and the 40 & over will be in Yakima August 23-25.  I've worked with a number of teams headed to Sectionals to provide Estimated Dynamic NTRP Rating Team Reports on both their own team and their opponents to help them plan line ups and strategy.  These reports include the current Estimated Dynamic NTRP Rating for all team members as well as their record on the team and full listing of what players played on each court and how many times.

Team reports aren't limited to the Pacific Northwest Section though.  I've provided team reports for teams in other sections including Hawaii, Northern California, Southwest, Inter-Mountain, Texas, Missouri Valley, Southern, Eastern, and the Caribbean.  Using these reports, one team has already advanced to Nationals, and several others are on their way to Sectionals.

It's not too late to get team reports.  Contact me and I can generate reports for you too.